History and principles of conductive media for standard DNA electrophoresis

ANALYTICAL BIOCHEMISTRY Analytical Biochemistry 333 (2004) 1–13 www.elsevier.com/locate/yabio

Review

History and principles of conductive media for standard DNA electrophoresis夽 Jonathan R. Brody and Scott E. Kern¤ Department of Oncology, The Sidney Kimmel Comprehensive Cancer Center at The Johns Hopkins University School of Medicine, Baltimore, MD 21231, USA Available online 28 July 2004

Abstract DNA electrophoresis has been a dominant technique in molecular biology for 30 years. The foundation for this common technique is based on a few simple electrochemical principles. Electrophoretic DNA separation borrowed from existing protein and RNA techniques developed in the 1950s and 1960s. For 30 years, common DNA electrophoretic conductive media remained largely unchanged, with Tris as the primary cation. DNA electrophoresis relies simply upon the negative charge of the phosphate backbone and the ability to distribute a voltage gradient in a sieving matrix. Nevertheless, the conductive properties in DNA electrophoresis are complicated by choices involving voltage, electric current, conductivity, temperature, and the concentration and identity of the ionic species present. DiVerences among the extant chemical recipes for common conductive media aVect central properties. Trisbased buVers, even in optimal form, create a runaway positive feedback loop between heat generation and retention, temperature, conductivity, and current. This is undesirable, leading to limitations on the permissible electric Weld and to impaired resolution. Recently, we developed low-molarity conductive media to mitigate this positive feedback loop. Such media allow for application of a higher electric Weld. Applications of DNA electrophoresis can now be reengineered for lower ionic strength, higher Weld strengths, and lower requirements for heat dissipation.  2004 Elsevier Inc. All rights reserved.

Molecular biologists must separate soluble nucleic acid molecules according to size. DNA electrophoresis thereby becomes an essential technique. Applications of DNA electrophoresis include analytic techniques such as restriction enzyme mapping, conWrmation of the identity of plasmid inserts and the products of polymerase chain reaction (PCR), sequence analysis, comparisons of polymorphisms among a population, and preparative techniques such as the separation of fragments for recovery and cloning and the quantitation of individual DNA species in a mixture. Practically every molecular biology publication today relies upon these techniques directly or indirectly.

夽 The authors are governed by the ConXict of Interest policies of the Johns Hopkins University School of Medicine. The authors' conXict of interest derives from rights they may have regarding intellectual property concerning new electrophoretic conductive media and as owners of Faster Better Media LLC, to which these rights have been assigned. ¤ Corresponding author. E-mail address: [email protected] (S.E. Kern).

0003-2697/$ - see front matter  2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2004.05.054

2

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

Electrophoresis now encompasses three major platforms of enclosing device: slab gel electrophoresis, capillary electrophoresis, and microfabricated devices. Slab gels, the most common form of DNA electrophoresis, involve molding a polymer (e.g., agarose) with conductive medium and applying a voltage so that multiple samples can migrate in parallel. Inversion and twodimensional arrangements of the electric Weld in slab gels are used for very large DNA fragments but command special equipment and are not entirely standardized. Capillary electrophoresis uses thin channels (usually of fused silica) to distinguish DNA fragments. The special features of capillary design are thus considered a specialized literature or industry trade secrets and will not be discussed here. Microfabricated/microchannel devices are relatively new; they separate DNA fragments as one feature of a more comprehensive analyte-processing system. The principles of electrophoretic separation of DNA by fragment size are largely unchanged by instrument design, but design-speciWc interactions at the interface of the enclosure and the matrix become increasingly important as the width of the path decreases. This article will primarily focus on conductive media in continuous electrical Welds, from the perspective of both reported and unpublished newer studies using DNA slab gels. The goal of a conductive medium is to maintain an orderly distribution of the electric Weld. This electric Weld is measured in volts/cm. The voltage gradient is established across a length of a gel. The properties of most immediate importance in separating DNA by size are the choice of sieving matrix, the pore size (its concentration), and the conductive medium used. The local strength of the electric Weld determines the movement of the ions in solution, producing both the current and the electrokinetic movement of the negatively charged DNA. The electromotive force acting on the DNA is equal to the local electric Weld (E) times the net charge (Q). This force is oVset by the friction of the DNA molecule against the conductive medium [1] and the sieving matrix. The basic principles for double-stranded DNA also hold for RNA and for single-stranded (denatured) DNA, although, for these, denaturants and heat are generally used to prevent intramolecular and intermolecular hybridization. Additionally, RNA is hydrolyzed by alkaline conditions, limiting the choice of conditions somewhat. RNA and denaturants are not covered in this article except in passing.

History The story of DNA electrophoresis is a 62-year history. In 1942, Coleman and Miller [2] reported the migration of neutral hexoses toward the anode with a borax (sodium borate) solution. Carbohydrate electrophoretic separation was more widely explored in the

1950s. Extensive and varied studies formed a set of rules for covalent borate complexation occurring with compounds containing adjacent –OH groups, such as carbohydrates [1,3]. For neutral sugars, high concentrations of borate provided the charge needed for successful electrophoretic separation of diVering sugars [4]. The Wrst electrophoretic separations of neutral carbohydrates were reported in 1952 [5]. Acidic carbohydrates were also assayed in nonborate buVers, but boric acid was often needed to provide separation. This separation depended on the formation of covalent borate complexes with the sugars [6,7]. Surprisingly, although this separation by voltage and chromatography could have directly set the stage for the use of conductive media to separate polynucleotide molecules in an electrophoretic device, it did not. Instead, RNA and DNA electrophoresis evolved with little change from the methods of protein electrophoresis. Most of the critical developments in DNA electrophoresis occurred in a brief spurt centered near 1971. Since this “golden era” of molecular biology there was no signiWcant progress in the evolution of the DNA electrophoretic conductive media for slab gel electrophoresis for nearly three decades. DNA was a common and notable contaminant in protein and RNA electrophoresis. For example, Thorne [8] in 1962 (a report comprising, in all, a single paragraph) separated poliovirus RNA from DNA, but this vignette is evident only in a reinterpretation provided by Thorne [9] in 1966. In 1965, Richards et al. [10] used Tris-based media to separate RNA and subsequent studies followed using similar methods [11]. In the early 1970s these electrophoretic techniques became reWned and solidiWed for applications to map the structure of DNA. Danna and Nathans [12] in 1971 used DNA electrophoresis to establish the relative molarity (number) and length of fragments of SV40 DNA that had been produced by the new restriction endonuclease from Hemophilus inXuenzae. Aaij and Borst [13] in 1972 used ethidium bromide in the conductive medium to illustrate how linear and circular DNA of various phage migrated diVerentially. At the time, they noted that the technique of DNA gel electrophoresis had “hardly been used.” Also in 1972, Hayward and Smith [14] separated and mapped the single-stranded DNA of T5 phage. Conductive media The exploration of diVerent ions in electrophoresis beneWted from diverse eVorts of many protein scientists over decades. DiVerent proteins have diVering charge densities, which vary with pH. This produced a great interest in the understanding of the leading and lagging edges of ionic zones, an interest that generally did not carry over into nucleic acid electrophoresis. For example, to isolate the property of “conductance” from the artifacts produced by mass ionic migration, protein

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

investigators cleverly used alternating current [15]. Various electrolytes included the use of Tris(2-amino-2(hydroxymethyl)-1,3-propanediol), acetate, borate, glycine, chloride, sulfate, and/or phosphate to control both pH and conductance. Their theories carried over into early thought with regard to RNA and DNA electrophoresis, although the utility of these complex rules in the newer Welds was perhaps more assumed than demonstrated. For example, in 1965 [10] it was argued that RNA mobility was intermediate between fast and trailing anions and should be focused in the large pore gel “to achieve adequate resolution.” Yet this report contained no related experimentation or discussion of the relative measured values of RNA and ion mobilities. Tris reigned For reasons not fully evident today, Tris became established as the favored cation for DNA electrophoresis. The loaded sample for RNA gels included EDTA [16], which often persisted in the derivative recipes for DNA conductive media, seemingly needlessly. Tris acetate (TA)1 was used in DNA electrophoresis in 1971 [12], where EDTA remained restricted to the loaded sample. Tris acetate EDTA (TAE) was used in 1972 as the gel conductive medium by two groups [13,14]. Tris boric acid EDTA (TBE) was used for RNA electrophoresis in 1968 [11] and for RNA sequencing in 1973 [17]. In the latter report, Maniatis (neither an author nor cited in the references) is credited in the text for having provided the key autoradiogram. His method is absent from the text other than mention of 7 M urea in the polyacrylamide gel. He was probably using Tris boric acid in some form. Also in 1973, Maniatis and Ptashne [18] used Tris boric acid MgCl2 to analyze protein–DNA interactions. Subsequently, DNA sequencing using TBE urea gels was used in a classic 1974 study employing base-speciWc chemical DNA cleavage [19]. Even though TBE and TAE dominated and emerged as the celebrity buVers in the Weld, other electrolytes competed. The early history of RNA electrophoretic buVers was derivative of earlier methodology. An example of this is found in a 1965 table [10] of anion/cation pairs that produced acceptable resolution. The authors speciWcally recommended acetate to be the counterion to use with creatine, cacodylate with imidazole, diethyl barbiturate with imidazole or with Tris, and glycyl glycine with ammediol. Barbiturate anion was common in early DNA electrophoresis, sometimes called by other names such as barbital or the trade name Veronal. Phosphate was not an uncommon anion for RNA in the 1960s and 1970s [20] and for DNA electrophoresis in the early

1 Abbreviations used: TA, Tris acetate; TAE, Tris acetate EDTA; TBE, Tris boric acid EDTA; MOPS, Morpholinopropanesulfonate.

3

1970s. A phosphate solution was recommended in the original version of the laboratory protocol book Molecular Cloning alongside TAE and TBE [21]. One early recipe included 36 mM Tris along with 30 mM sodium phosphate; the conductance must have been exceptionally high in this case [14]. Chloride anion was the most common to accompany the Tris cation in biologic buVers and often did so in protein electrophoresis, but only occasionally was Tris–HCl explored in DNA electrophoresis [13,14]. Saline has been largely avoided in DNA electrophoresis. The use of sodium or, alternately, chloride in any appreciable molarity is usually exclusive of the other ion. Exceptions are found among the earlier studies [22,23], which usually concerned low-resolution electrophoretic measurements of DNA mobility (often in solution rather than in gels). Like the sodium borate methods to separate neutral sugars from the 1940s, such sodium-based methods were not adapted for high-resolution separation and mapping of multiple DNA fragments in gels; they were, however, used for high-resolution RNA gels, where sodium morpholinopropanesulfonate (MOPS), an anionic biological buVer amine) became commonplace. Sodium dodecyl sulfate (SDS) was sometimes included in early DNA electrophoresis, perhaps to prevent undesirable interactions of DNA with contaminants in the sample and the matrix [12–14]. Yet even with these contrarian explorations, TAE and TBE remained predominant. Evidence of this is betrayed in the newer laboratory protocol books that recite only Tris-based media [24,25]. Samples and denaturants Sample-loading solutions (“loading buVers”) are of two major categories: (1) those for nondenaturing gels and having ionic conditions roughly matched to the gel and (2) those employing denaturing conditions, the solution often being deionized, and incorporating no attempt to match the gel ionic conditions. Other ingredients of these loading solutions include dyes, detergents such as SDS or sodium sarcosine, EDTA to reduce intermolecular interactions mediated by divalent cations, dense and viscous additives such as Wcoll, sucrose, and glycerol to enable settling of the sample in or on the gel, and RNase to reduce the size of RNA fragments that are often copuriWed with DNA. EDTA in the sample medium prevents the denaturation of DNA that is otherwise facilitated by dyes at low-ionic strength [26], although the sodium content of the EDTA must be responsible in part for this property [27]. These ingredients can add undesirable properties. Glycerol produces “streaming” of the sample to yield U-shaped bands [27] and can react with boric acid ion in the gel to disturb resolution [28]. The counterion that accompanies the dyes or detergents can produce bending of the bands in individual lanes

4

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

enhancing of migration distances or a simple loss of resolution. Excess salt can be mitigated by diVusion when a sample is allowed to “sit” in the well for a time before separation [29]. Newer alternatives to the laborious loading of liquid samples include Wlters or solid matrices that absorb or bind the DNA until applied directly and very eYciently to the gel [30], termed “combs” in one popular manifestation. Denaturing conditions do not as a rule aVect the fundamental principles of electrokinetic movement of the nucleic acids. To achieve denaturing conditions, the gelloading conditions can incorporate formamide, urea, heating, and alkaline pH for DNA gels. For RNA, the most common denaturant is formaldehyde. Although RNA and DNA hybridization is promoted by higher salt concentrations, little attention has been devoted to improving the electrophoretic resolution of denatured polynucleic acids through the reduction of ionic strength of the conductive solution. This seems a valuable direction for future research.

To the heat of the solution The diYculties encountered in the development of protein electrophoresis accent the key importance of joule heating. Two major problems were overcome. In free boundary electrophoresis, the thermal convection produced by heating was initially managed by running the separations near 4 °C, this being the point of maximal density of the conductive media, to minimize the disturbance created. Later, this problem was managed by zone electrophoresis, which took place within a matrix. The second problem was the quantity of heat retained by the separative cell. This was in part managed by changes in the shape and environment of the cell or separative support (such as Wlter paper) to allow more eYcient heat radiation [15,31]. Interestingly, there was apparently at times a misconception that eVective protein separation had a “need for a strong current” [32]. Although agar was used in 1951 for electrophoretic separation of nucleic acid components, the nucleic acid Weld soon found a preference for paper as the conductive matrix, for with agar “it is diYcult to avoid overheating and consequent ‘sweating' of the gel” [1]. Nucleic acid and protein researchers in the early 1950s often immersed the paper in a bath of the nonpolar insulator carbon tetrachloride or chlorobenzene to achieve cooling [1,4]. In the early days of nucleotide electrophoresis, 50 mM ammonium formate or ammonium acetate was often preferred to allow the use of higher voltage; in contrast, the higher conductivity of 50 mM phosphate or borate solutions necessitated a reduction of the voltage gradient [1]. Eventually, protein separations were cooled by the use of gels sandwiched between two glass plates or thin plastic membranes [4].

New designs marketed for DNA electrophoresis hearken to the 1950s practices in protein electrophoresis. The most recent products include thinner slab gels and capillaries, a reduction of the cross-sectional path for electric current by a limitation or elimination of the older practice of gel immersion in a reservoir of conductive medium, and the encasement of slab gels between two plates of thin plastic. Melting of the gel still occurs at low-voltage with the newest agarose platforms. This is not unexpected, for no fundamentally new engineering properties are incorporated in the gels. A runaway positive feedback loop and new explorations Heat generation sets limitations for gel electrophoresis. Heating restricts the amount of voltage applied to the gel system, since it can denature the sample or reduce gel integrity [33,34]. Heating is explained in part by Ohm's law and the power law where voltage (V), current (I), and power (P) are directly related (P D VI) [33–35]. These variables are ultimately aVected by ionic conductance due to choice of the solutes and their concentrations in the conductive media. Heating is also dependent upon the proportional relationship of temperature and the conductivity of a solution. This latter relationship is implied in a frustration expressed in a report of carbohydrate electrophoresis from 1952 [5]. Although the voltage was held steady, “the temperature rose a few degrees during the run. The current usually increased by 30–40 per cent.” This frustration applied also to DNA electrophoresis. We found that exploration of the central electrothermal relationships led to promising new approaches in electrophoretic conditions. At constant voltage, a positive feedback loop existed between the temperature of the buVer and the current for all Tris-based conductive media (Figs. 1A–C) [35]. At constant voltage, TAE and TBE gels experienced an increased temperature and current over time (Figs. 1A and B). To verify that temperature and current were directly interrelated, we manipulated the temperature of the media. For TAE and TBE, external control of the temperature of the conductive media resulted in increased current (Fig. 1C). SuperXuous sodium ions and the high amount of Tris ions in these buVers were responsible for producing unnecessary current and limited the ability to run gels at a high voltage. Hence, in these common conductive media, heat and current were related in a positive feedback loop that limited the permissible voltage (Fig. 2). This principle lays the foundation for the electrothermal properties of DNA electrophoresis in which temperature directly aVects the conductance of the electrophoretic system; the rate of heat generation accelerates over time, placing strict limits on the speed of electrophoretic separations. We explored the constituents of TBE and TAE as prepared by major biotechnology suppliers, laboratories within our institution, and standard laboratory

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

5

Fig. 1. EVects of conductive media on current and temperature during agarose electrophoresis. TBE is 89 mM Tris boric acid, 2 mM disodium EDTA; TAE is 40 mM Tris acetate, 2 mM disodium EDTA [24]. “A,” 10 mM sodium acetate, 2.5 mM NaCl; “B,” 10 mM sodium boric acid pH 8.5 [35]. SimpliWed EDTA-free versions of Tris boric acid (20 mM Tris, 20 mM boric acid) and Tris acetate (30 mM Tris, 20 mM acetate) were prepared (A) Heat generation by conductive media over time during electrophoresis at constant voltage (150 V) in a closed electrophoretic system. Media were not recirculated or cooled. At longer time points, temperature continued to increase at similar relative rates. Using simpliWed Tris boric acid, temperature did not change; using simpliWed Tris acetate temperature rose 4 °C. (B) Conductance of media, measured in the same experiment as (A), over time at constant voltage (150 V). With use of a simpliWed Tris boric acid, current increased from 18 to 19 mA; with a simpliWed Tris acetate, current increased from 100 to 111 mA. (C) EVect of external heating of conductive media on conductance. Solutions were heated or cooled to 17, 24, and 30 °C and the current was immediately measured. (D) Equal concentrations (2 mM) of diVerent forms of EDTA used in electrophoretic media were added to TAE (40 mM Tris) and compared at constant voltage (150 V) during electrophoresis. Lines connect the measured data points (Wlled circles). Studies of conductance, temperature, and voltage used a horizontal gel rig with safety cover (MGU-500, CBS, Del Mar, CA) and a power source that provided current readings at set voltages (FB 570, Fisher Biotech, Pittsburg, PA). All media were analyzed at the same volume (650 ml) in the reservoir.

Fig. 2. Fundamental electrothermal relationship of DNA electrophoresis. The “runaway” positive feedback loop is created by an interactive increase in temperature and conductance of the conductive media during electrophoresis.

manuals (such as Molecular Cloning) [24,36]. Various protocols use either free acid EDTA (Invitrogen, Carlsbad, CA), disodium EDTA (Current Protocols in Molecular Biology) [24,36], or tetrasodium EDTA (USB, Cleveland, OH). Although this might appear to be a trivial diVerence, media containing equimolar EDTA of diVerent forms behaved quite diVerently. As expected, media containing EDTA (acid) had a lesser

current initially and maintained a reduced current over time, as compared to the two other forms of EDTA that contained additional sodium (Fig. 1D). We concluded that amounts of sodium capable of generating signiWcant current are included in common media. Another protocol speciWed sodium acetate to be added to TAE along with disodium EDTA, perhaps a holdover from the use of sodium acetate in conductive buVers for RNA gels in the 1960s [14,16,37]. EDTA, introduced initially into the sample buVer to prevent RNA from remaining at the origin [16] and to protect against endonuclease activity, is now superXuous, since most DNA samples are readily soluble and since commonly used enzymes today would not carry an undesirable enzymatic activity under electrophoretic conditions [16]. We also found that EDTA was not needed even when using electrophoretic media based on creek water (Pinelands, NJ) and tap water (Baltimore, MD) (data not shown), presumably ruling out a general need for inhibition of endonuclease activity. The addition of free acid EDTA into the conductive media would be minimally conductive, however, and could be used in situations where EDTA may be speciWcally desired. Another ionic source is also often overlooked; ammonium persulfate,

6

present in polyacrylamide considerable current.

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

gels,

contributes

Caveats for Tris- and acetate-based electrophoresis As judged by the reduced current, temperature generation, and gel resolution, TBE was a better conductive medium than TAE (Figs. 1A–C, and 4). It is not unusual to encounter a 0.5£ form of TBE, having 45–50 mM Tris [24] or a 0.5£ TAE, having 20 mM Tris in particular applications, such as in analysis of protein–DNA interactions or in electrophoresis of large DNA under refrigerated conditions. The underlying reasons for use of a particular medium are not as a rule well documented, but one can take advantage of the reduced joule heating of dilute solutions and of the improved separation of large DNA fragments by acetate ion. In continuous voltage electrophoresis, acetate separates large DNA fragments better than borate, although borate resolves smaller DNA fragments (2 kb and lower) well. This has led some investigators to favor TAE in the resolution of larger DNA, and to favor TBE for smaller fragments. This diVerence among the conductive media may be due to borate–DNA complexes and is partially mitigated by large (greater than 45 mM) concentrations of borate [38]. These guidelines do not hold for Weld inversion gel systems, where borate systems can separate megabase DNA fragments admirably. SimpliWed versions of common Tris-based media We explored simpliWed versions of TBE and TAE, some lacking EDTA and sodium, at lower concentrations of Tris; this approach provided limited practical beneWt. These media could mitigate the feedback loop but had impaired resolution when used at higher voltages and when using Tris at less than 30 mM (data not shown and Fig. 1) [39]. Indeed, Loening [40] in 1969 explored a “low-salt buVer” that included 30 mM Tris– HCl.

Alternatives to Tris acetate and Tris boric acid Conventional Tris media have several additional disadvantages, including high cost, precipitation of stock solutions (speciWcally TBE), and recipe complexity. Due to the disadvantages of TAE and TBE, others and we explored new media. For example, a variety of alternate electrolytes for DNA electrophoresis was explored in the early years. Some of this history was discussed above. Once TAE and TBE became established as the default conductive media in the mid-1970s, alternative ions were still occasionally explored, but the use of alternate conductive media for slab gels was usually experimental or temporary and never widespread. The exploration of

alternative cations and anions is well developed in capillary electrophoresis, although relatively few of these alternatives have been evaluated substantially for DNA separations [41]. The acceptance of alternate ions was limited by their own high costs, low availability, a dearth of evidence for their signiWcant advantages, and a progressive standardization of common molecular biology protocols [21,25]. There was some opinion favorable to Tris that, although amines and borate bound to DNA, Tris did not [38]. There were also concerns that alternate ions might produce unpredictable results in an otherwise familiar system. For example, while pore size was unchanged by the choice of conductive medium in which the gel was cast, it could appear to change were electroendosmosis not taken into account [42]. Further concerns remained that certain matrices required a threshold ionic environment to retain integrity. Rules governing such interactions are not generalizable and will not be covered in this article. Despite the general lack of enthusiasm for alternate conductive media, the choice of conductive media for DNA electrophoresis can be used to lower the current per unit voltage, resulting in lower heat generation and better performance [35,39]. Investigators searching for alternative media in capillary electrophoresis have had some success [41]. There is justiWcation for additional exploration of alternate media. Amine cations serving as alternates to Tris have been suggested. In an interesting suggestion, the matching of pK values of the cations and anions was proposed, as represented in the ionic pairs of ethanolamine:Capso pH 9.6, triethanolamine:tricine pH 7.9, and bis Tris:Aces, pH 6.7. These new media outperformed TAE and TBE and appeared to represent a promising new direction. The report, however, was not speciWc about the relative concentrations of sodium ions in each solution tested, a detail that could have dominated the performance diVerences observed [39]. Other cations (morpholine and imidazol) were also explored to replace Tris, but resolution suVered [43]. The metal cation sodium is standard in RNA electrophoresis in the form of 7–9 mM sodium MOPS. Metal cations are nearly unexplored in the literature of high-resolution DNA electrophoresis. In other forms of electrophoresis, for example, univalent metal cations are known to alter the electrophoretic properties due to their binding to surfaces of micelles and detergents [44], properties with essentially no current relevance to the electrophoresis of DNA. Divalent cations such as Mg2+ have been used to stabilize protein–DNA interactions during electrophoretic assays of dissociation rates [18], but their capacities for precipitation and for crusting on surfaces have made them no more than occasional constituents in the conductive media. We thus explored monovalent metal cations, as guided empirically.

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

We also explored organic and inorganic anions, Wrst comparing acetate and borate (Brody et al., unpublished). Some alternate anions from the literature were noted above. Interestingly, MOPS is successful as a medium perhaps because it contains small amounts of sodium and no Tris and thus no superXuous ions. To reengineer newer electrophoretic media in a manner perhaps more free of conventional design biases, we chose to systematically explore unbuVered mixtures of salts comprising anions of high and low electromobility (electromobility is discussed below). Combinations of sodium chloride and sodium acetate provided satisfactory electrophoretic performance near 10 mM total sodium ion concentration (Fig. 3). Importantly, these studies identiWed ranges in which sodium served as an acceptable cation. Sodium chloride has a known ability to disrupt complexes of DNA with itself or with other ions [45], a property that may contribute to our observation of optimal composite media. Sodium chloride/acetate media thus provided a new paradigm from which to design a novel and distinct family of low-conductive metal cation conductive media for DNA electrophoresis. A sodium chloride/acetate combination conducting the least current and having excellent resolution over a prolonged run time was chosen as an working example of this form of medium (Figs. 1A and B, 4A and B). In contrast, due to rapid ion exhaustion, media containing predominantly sodium chloride produced in our

7

studies in slab gels a phenomenon that we interpret as a salt boundary (a form of “stacking;” Fig. 3), which was associated with a moving zone of deformity in the thickness of the gel. Low-molarity sodium chloride was an adequate medium for resolution for DNA electrophoresis with some design alterations, however, it did not mitigate the positive feedback loop Figs. 1A and B. In our explorations, we were able to prevent our salt boundaries with a simple increase in the eVective volume of the anodal reservoir or by recirculation of the solutions between the two reservoirs (data not shown). This demonstrated that the problem was one of ion exhaustion and not one speciWc to altered ion migration at the reservoir/polymer interface. A similar problem may have been described in 1951 by Gordon et al. [46], who successfully addressed the anodal swelling of the gel by an increase in the ionic strength of the anodal reservoir of electrolyte. A related problem in polyacrylamide gels was described by Spencer [47] in 1983 and by others who noted anomalous conductivity zones that developed during electrophoresis at the cathodal end of electrophoretic capillaries and by Bilenko et al. [48] in 2003 who noted anodal resistive zones in capillaries. These deleterious zones were attributed by Spencer [47] and followers [48] to a diVerence in ionic transfer number at the reservoir/polymer interface. Figeys et al. [49] were able to overcome this ionic depletion by increasing the salt concentration of the reservoir but adhered to the basic

Fig. 3. Salt boundary in sodium acetate/sodium chloride agarose gels. (A) 11 mM sodium chloride, 1.5 mM sodium acetate run at standard electrophoretic conditions (8 V/cm, 11 h). (B) 10 mM sodium acetate, 2.5 mM sodium chloride run at standard electrophoretic conditions (8 V/cm, 11 h). A ladder (1 kb Plus; Invitrogen, Carlsbad, CA) and an unpuriWed PCR product in reaction buVer were applied using glycerol loading solutions.

8

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

rose (Brody et al. unpublished). Sodium boric acid performed as well as conventional media, TBE and TAE, at conventional temperature (21 °C) and voltage (8 V/cm) in the size range of 100–2000 bp, conditions under which sodium boric acid could be run at three or more times faster than TAE or TBE due to its lower Joule heating, permitting a much higher voltage [35]. Sodium boric acid also performed well in polyacrylamide gels under denaturing conditions (155 °C, urea) (data not shown). Sodium boric acid as the conductive medium was able to abrogate the “runaway” feedback loop of increasing current and temperature (Figs. 1A–C and 2). Sodium chloride/acetate performed better than TAE in analysis of the feedback loop (Figs. 1A–C) yet was unable to mitigate the feedback loop to the same extent as did sodium boric acid. pH

Fig. 4. pH changes in DNA electrophoresis. “A,” an acetate-based buVerless conductive medium as described in Fig. 1 (10 mM sodium acetate, 2.5 mM sodium chloride) that produced resolution similar to TAE; “B,” 10 mM sodium boric acid pH 8.5 [35]. (A) “BuVer” exhaustion. pH changes of conductive media were analyzed in the cathodic chamber during electrophoresis (10 V/cm, 1 h). (B) pH changes of conductive media were analyzed in the anodic chamber during electrophoresis (10 V/cm, 1 h). See methods in Fig. 1.

theory of Spencer [47]. In any respect, a beneWt for including a nonchloride anion in addition to any optional chloride ions arose in our explorations, for these anions prevented the salt boundary without resort to larger reservoir size or recirculation. That is, in simple rig designs, low-molarity sodium chloride media proved insuYcient as an electrolyte for DNA electrophoresis. In contrast, we did not observe salt boundaries when a nonchloride anion was substituted (acetate at greater than 2.5 mM, for example, suYced; Fig. 3). Brief studies using polyacrylamide appeared to give results similar to those using agarose (data not shown). These explorations allowed us to reanalyze the common perception of Tris as “the best” cation for DNA electrophoresis. Sodium chloride/acetate media could be used as starting points for anion substitutions. A clearly exceptional anion was borate. Sodium boric acid at 10 mM sodium was excellent at almost all electrophoretic conditions [35]. Variants of sodium boric acid worked best within a range of 7.5 and 12.5 mM sodium when used with DNA fragment sizes of 100–2000 bp in conventional concentrations of agarose gel (0.5–2.0%). At a 2.0% agarose concentration, sodium boric acid at this and at much lower concentrations could resolve very small fragments of DNA (20–100 bp) successfully in aga-

The pK values of the functional groups of the nucleic acids were established in 1925 and 1926 [1,50–52]. The primary phosphate groups of nucleotides are fully dissociated below pH 2, while the amino groups of the bases dissociate between pH 2 and 5 and allow diVerentiation between the mononucleotides in electrophoresis at acid pH. The secondary phosphates have essentially identical pK values at 6. The enol groups Wrst to dissociate have pK values near or above 9 [1]. The electrophoresis of polynucleotides thus solely depends upon the charge of the ionized linking phosphate groups when separations are performed at a pH range of 5–10. DNA molecules migrate most rapidly (have the largest negative net charge) above pH 7, conditions at which, for practical purposes, the phosphates are fully deprotonated. Within these considerations, the pH is not a major factor for electrophoresis of DNA or RNA [10]. Most researchers refer to conductive media as “buVers.” Nonetheless, a more circumspect treatment is valuable. Tris and borate are electrophoretic buVers, but acetate (pKa of 4.7) is not. We conWrmed these expectations by using pH monitoring to capture the changes in pH in the cathodic and anodic chambers of a rig over the course of an electrophoretic run (Fig. 4). It is not well known that unbuVered solutions, such as sodium chloride/acetate (Figs. 1, 3, and 4) or lithium acetate (Brody et al., unpublised), could serve as acceptable solutions for DNA electrophoresis. General principles of electromobility and conductivity The conductivity of a material is the reciprocal of its resistivity, an innate property of a given solution or material independent of thickness or length. These properties can be distinguished from the total conductance or its reciprocal, the resistance, of a quantity of a material. The conductance depends not only on the conductivity

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

but also on the length of the current path and the crosssectional area orthogonal to the current. The conductivity of an electrolyte in solution is the product of its concentration and molar conductivity. Molar conductivity of a dissociable solute increases with ionization and thus depends on dilution and pH; it is decreased by any interaction with other solutes and surfaces. It is easy to overlook that the cations and the anions contribute independently to conductivity and that one or both can become locally depleted (termed electrolyte exhaustion) during the course of electrophoresis [35]. The transport number is the fraction of the current carried by each ionic species. Molar conductivity depends, in opposing ways, upon both the charge of an ion and the friction (viscosity) that it encounters as it passes through a solvent. Ions do not exist alone but are accompanied by a shell of solvent and sometimes by other ions. Thus, if ions of two types are of a similar atomic dimension and charge but one has a larger solvated size, the latter will be less electromobile and less conductive. Lithium, for example, is less conductive than sodium. Cation and anion mobilities are known. The measured cationic mobilities, relative to potassium ion, are as follows: ammonium 1.00, sodium 0.68, Tris 0.40. Near its pK of 8.3, Tris was found to be reversibly protonated and thus proportionally less electromobile than this value would indicate [53,54]. Others have measured anionic mobilities, again relative to cationic potassium, which follow: acetate 0.56, chloride 1.03, monobasic phosphate 0.45, dibasic phosphate 0.39, propionate 0.49, sulfate 0.55, borate 0.54 [53–55]. In agreement with these expectations, we found that a 20 mM Tris–HCl solution at pH 8.1, having somewhat more than 10 mM each cations and anions, has 73% of the conductivity of a 10 mM NaCl solution (and a 10 mM sodium borate solution has 48%). Consequently, Tris media usually incorporate 40– 80 mM Tris. The discrepancy in the necessary concentrations of Tris and sodium ions in acceptable forms of electrophoretic conductive media is thereby rationalized. The literature, despite its inconsistencies with regard to the preferred constituents of a conductive medium, contains repeated warnings that highly conductive ions are to be avoided. In the early 1950s, it was noted that the use of borate or phosphate anions, in place of formate or acetate, caused excessive heating in nucleic acid electrophoresis. In an early paper on RNA electrophoresis [37], Tris base and diethylbarbituric acid were used, “free acid and free base being chosen to avoid ions of high mobility.” It is rather clear how the data on conductivity can be used to engineer some desirable modiWcations into DNA electrophoresis. Low-conductive solutions would permit a reduced heat generation and permit higher voltage, speeding the electrophoretic separation of DNA and reducing both the real and the apparent causes of poor resolution. In our most recent explorations, we took advantage of the low-conductivity

9

of lithium ion to design boric-acid-based and acetatebased media of even lower conductivity (Brody et al. unpublished). Lithium ion has a larger shell of hydration than does sodium and consequently a lower electromobility, and for small DNA fragments (20–100 bp), various boric acid media incorporating alkali metals or ethanolamine as the cation could be used in 1 mM solutions for rapid, high-resolution, ultra-high voltage (up to 150 V/cm) separations in slab agarose gels (Brody et al., unpublished). Additional comments on boric acid and boric acid ion Boric acid has the formula H3BO3 or B(OH)3. Its anion, B(OH)¡4 , unlike other oxyanions, is formed by the addition of a hydroxyl group. Despite its multiple hydrogen atoms, boric acid is monobasic. The anion is termed the boric acid ion, or simply borate. The pK is variably reported from 9.1 to 9.4. At or below this pH, a boric acid solution serves as an eVective buVer against the production of hydroxide at the cathode. Boric acid anion migrates anodally. At the anode, it can donate a hydroxyl ion to buVer the production of hydronium ion. At a standard electrophoretic pH range of 8.0–8.5, the vast majority of boron is in the form of uncharged boric acid and will have no electromobility in that form. Because electrophoretic solutions do not predominantly contain the anion but instead contain uncharged boric acid, they can be more properly termed with names such as “sodium boric acid” or “Tris boric acid.” The lower the pH of the conductive medium, the larger the fraction that remains as boric acid, a reserve for the buVering of the electrode reactions and for “replacing” the ionic migration toward the anode [35]. Boric acid is also oxidized and presumably cannot be chemically destroyed at the anode, unlike acetate. These properties account for the superior performance of boric acid solutions to delay both electrolyte exhaustion and buVer exhaustion, as demonstrated above (see Fig. 4) [35]. A more accurate explanation would convey that boric acid and its anion are in a dynamic equilibrium, governed by the pH of the solution. A given boron atom migrates in an electric Weld with a speed proportional to the ionized fraction and thus relatively slowly compared to anions of carboxylic acids. Boric acid ion has long been known to form a covalent complex with adjacent cis-hydroxyl and, although perhaps less avidly, with other vicinal dihydroxyl groups [56]. This accounted for the usefulness of boric acid solutions to allow separation of neutral carbohydrates [3]. Boric acid ion, having four oxygens, when present in limiting concentrations, is reported to form complexes (type III complexes) that link diVerent carbohydrate molecules [1,56]. For this reason, some reports have advised large concentrations of boric acid for the purpose of saturating each molecule of polymer, precluding cross-link-

10

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

ing. One might think that these considerations could potentially complicate separations of RNA, but not necessarily those of DNA, as DNA lacks the cis-hydroxyls found in ribose. Nonetheless, there is experimental evidence for complexation of DNA in boric acid solutions. For example, DNA complexes were detected readily with use of 0.5£ TBE (45 mM Tris) as compared to standard TBE (90 mM Tris) [57]. Early studies [56] had found evidence for such complexation in certain nonplanar orientations despite the lack of vicinal dihydroxyls, which presumably must be provided in DNA. Such considerations have encouraged the use of high ionic strengths to achieve optimal DNA resolution. The exact nature of DNA–borate complex remains uncertain, and they are not always detected. For example, such complexes are not always observed in agarose gels, and a competition of agarose for boric acid ion was suggested as the reason for their absence [57]. The anion also forms complexes with glycerol, which might have some signiWcance for loading media that contain both. Boric acid solutions partially inhibit the interaction of DNA with silica surfaces when in the presence of chaotropic salts, such as used in the glassmilk procedure [58]. Lower concentrations of boric acid, such as present in 0.5£ TBE, sodium boric acid medium [35], and acetatebased media, readily permit fragment isolation and cloning from agarose gel slices.

Caveats for all conductive media The voltage applied by the power source greatly exceeds that applied to slab gels. Critical electrical properties are best measured at or on the gel. In all buVered systems tested by us in slab gels, the measured voltage “seen” by the gel (using voltmeter probes and measuring across the length of the gel) was roughly half of the setting on the power supply when tested across the gel. Further, the potential diVerences across the “upper” and “lower” halves of the gel were equivalent and stable over the course of a standard electrophoretic run (data not shown). The electrophoretic force experienced by each segment of the gel thus was consistent with the concept of Weld strength, calculated by dividing volts by distance, although this would be modiWed locally by the exhaustion of individual ionic species through migration [13,36]. Most of the remaining voltage drop and power consumption is assumed to be attributable to the electrolysis of water. This is justiWed by the known high energy consumption required for the rapid electrolysis that occurs during electrophoresis, which by Ohm's law commands a voltage step-down (potential diVerence) at the site of the reaction. The reservoirs that electrically link the electrodes and the ends of the gel characteristically have a high orthogonal cross-sectional area as compared to the gel and thus account for relatively little

voltage drop. The proportion of voltage drop attributable to electrolysis would thus be relatively higher for gels of greater cross-sectional area or of shorter length. The values of the electric Weld in the literature, unfortunately, usually specify the voltage delivered, rather than that experienced by the gel [24]. It is, along these lines, possible to cool the gel by immersion in a less conductive reservoir solution. This was intentionally applied by the use of less conductive solutions such as reservoirs of organic solvents in early protein electrophoresis or as likely achieved by Kozulik [59] to permit a reduction in band slanting (discussed below). The use of a low-conductance loading medium can be suggested to elevate the initial electric Weld across the sample and is expected to produce tighter bands. This is an inherent practice in denaturing DNA electrophoresis when the loading medium includes deionized formamide. The cross-sectional area, measured perpendicular to the direction of current, is a determinant of resistance and inXuences the voltage drop that occurs along the length of a gel. The generous width of the reservoir chambers at the electrodes oVers a very high conductance, rendering the distance of the electrode from the gel a minor consideration. In contrast, wedge gels have a gradient in electric Weld strength, with a lesser Weld and slower speed of DNA migration where the gel is most thick. Wedge shapes were developed for sequencing gels to progressively retain smaller fragments that otherwise would migrate into the anodal electrolyte reservoir [60]. It is possible to reduce the overall voltage requirements minimally by placement of the electrodes nearer to the gel, but this would not directly change the heat generation within the gel. Low-concentration conductive media do not necessarily “exhaust” faster in our explorations, a favorable Wnding that is attributable to a lower current, which compensates for the reduction in the ionic reserves [35]. Ion exhaustion is reduced in boric acid media as compared to acetate media, due to the reserve provided by boric acid [35]; i.e., its average electromobility is reduced. Redox reactions at the electrodes are often overlooked The cathode forms hydrogen gas and hydroxide ions, and the anode creates oxygen and hydronium ions, by the energy-consuming electrolytic decomposition of water discussed above. Both gasses are diatomic molecules. At times when a question of the correct polarity in an equipment setup may exist, it is generally easy to distinguish the anode from the cathode by the twofold greater rate of bubbling at the cathode. Other expected reactions include the reduction of sodium ions at the cathode to form nonionic sodium, which immediately reacts with water to generate sodium hydroxide and hydrogen gas. Other anodic reactions can also be suspected. In adopting TAE for nucleic acid

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

electrophoresis, Loening and Ingle [61] chose acetate instead of chloride since chloride would be expected to produce hypochlorite by electrolysis. Radicals are formed by oxidative decarboxylation at the anode. In Tris media, a peracid derivative of Tris formed at the anode. Indeed, the anode generated an oxidant species for all compounds tested that contained hydroxyl or carboxyl groups [62]. The literature on the redox reactions can be confusing. For example, an insightful report [39] suggested a rational approach for designing alternate conductive media, yet the published mechanistic illustration of the redox reaction at the anode depicted the loss of an electron as literally removing the charge of an anion. In fact, such a loss would chemically destroy the anion. A redox reaction had been confused for an acid–base reaction by the illustrator (the anode generates hydronium ions, the representation of which would have saved the Wgure). The substitution of sodium salicylate for the sodium acetate in sodium chloride/acetate medium illustrates this reaction class. The anodic chamber becomes bright yellow due to the oxidative decarboxylation of salicylate followed by oxidative meta-cleavage of the catechol intermediate to form 2-hydroxy muconic semialdehyde. This reaction is duplicated by leaving a bottle of sodium salicylate solution open over time; the solution similarly turns bright yellow (data not shown). The chemical eVects of the redox reactions would be critical for deciding whether and how long to recirculate the medium during electrophoresis or whether to reuse media in multiple runs. Unfortunately, the consumption rate of the ions is not generally known and is certainly not considered in formulating such decisions in individual laboratories. More caveats on heat and resolution DNA migrates faster under high temperatures [63]. This accounts for the “smiling” of DNA bands across a gel. Similarly, the narrowing of a lane in a gel indicates a relatively high salt concentration in the loaded sample. Higher temperatures increase the diVusion of DNA, causing a loss of resolution. In slab gels, temperatures are better controlled with vertical rigs, translating into improved resolution. In contrast, horizontal gels experience a considerable problem; they have an overlay of conductive medium which can increase the temperature of the top of the gel. This creates the appearance of a lower resolution in horizontal gels. Yet, cutting the gel along the lane shows that this is in part a visual illusion produced by the bands slanting by means of their migrating faster near the top surface of the gel than at the bottom [59]. Cooling of both faces of a gel thus can improve the real and the apparent resolution. Gels with

11

less heat generation or having more eVective heat radiation oVer better resolution. It is also possible to use vigorous convection (which can be achieved by recirculation of the medium) or to use a lower-conductive reservoir solution (noted above). Such measures to mitigate the eVects of Joule heating especially can aid the resolution of smaller fragments, due to the severity of the problems of diVusion and of band slanting being inversely proportional to the fragment length. An increased volume of buVer in the gel rig increases both the current and the temperature during electrophoresis at constant voltage. Some investigators regretably believe that the use of a greater depth of overlay of conductive medium would act as a heat sink for the gel. New precast designs now reduce the thickness of the gel, and some designs eliminate the overlay usually found in horizontal gels to accommodate the limitations of Tris media. In the past, vertical rigs were usually used to accomplish similar goals.

DNA electrophoresis revisted One of the two invariant components of commonly used conductive media, EDTA, remains unexplained [64]. The other, Tris, is ubiquitous in homemade gels, in precast slab gel systems, and in most capillary systems. But the 30-year dominance of TAE and TBE is being seriously challenged. It is worthwhile to consider the conditions that accounted for the acceptance and persistence of these chemical solutions. As mentioned earlier, there was a progressive standardization of the common published laboratory protocols. But while this would explain the adoption of Tris by the masses of researchers, the continued acceptance by the most insightful investigators would not be explained. Other explanations are required. First, a familiarity with the older conductive media could account for this stagnation. Choices among protein electrophoretic buVers had been heavily constrained by the diYculty in managing the physical properties of proteins, a diverse analyte. It may have been assumed by the early investigators of nucleic acid electrophoresis that an improvement upon these buVers would oVer rewards only with a considerable investment of time and attention. Indeed, Richards et al. [10] in 1965 argued for a governing theory of protein-like complexity and that RNA mobility was intermediate between fast and trailing anions and should be focused in the large pore gel ”to achieve adequate resolution.” Given the excitement oVered by the alternative uses of their eVorts (the mapping and sequencing of the Wrst genomes), the familiarity with the protein buVers contributed a contentedness; these solutions found unmodiWed adoption for nucleic acid electrophoretic separation by even the most creative investigators.

12

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13

Second, a “Catch 22” also accounted for quick acceptance of Tris media 30 years ago and the lack of signiWcant further maturation of the Weld. Investigators would have beneWted from a change from Tris to more electromobile cations, but such a simple change produced excessive current at similar concentrations. Also, investigators would beneWt from reducing the ionic strength, but in a Tris system this led to poor resolution (i.e., below 0.5£ TBE). The solution lay in the simultaneous exploration and change of both factors. Third, the concern that DNA complexes would form and result in reduced resolution, in conductive media having limiting concentrations of boric acid ion, supported the continued use of boric acid solutions in which the anionic form was present at over 30 mM. In such solutions, the nonionic form was present at many times this value. Finally, there has been a peculiar orthodoxy to adhere strictly to the term “buVer.” When explaining the electrical advantages of new conductive media to otherwise capable colleagues, the common questions are “What provides the buVering?” or “How well does it buVer the pH?” A better Wrst question would be “How well does it distribute the electric Weld over a given period of time?” Perhaps investigators' perceptions have been channeled inappropriately by use of the term “buVer” for all conductive media. It is possible that the 30-year arrested maturation of conductive media was responsible for a signiWcant loss of time and money. Investigators who commonly use slab gel electrophoresis can readily move to a more successful system of low-ionic, low-conductive media such as sodium boric acid [35]. We recently estimated that for agarose electrophoresis alone in the United States, the scientiWc community would have saved over 30 million dollars last year in reduced chemical expenditures [35]. If one includes the cost of materials, labor, time to make up solutions, etc. as “total process cost,” one can estimate a potential savings of nearly 35 dollars for each hour recovered. With such considerations in mind, it is interesting to imagine a fantastic history in which Tris-less media were adopted in the early 1970s. For example, at what earlier date would the human genome project have been accomplished and at what fraction of the cost?

Note added in proof The cited unpublished observations of J.R. Brody are now presented in J.R. Brody et al., BioTechniques (2004, in press).

Acknowledgments This work was supported by NIH Grant CA62924.

References [1] J. Smith, Electrophoretic separation of nucleic acid components, 1 (1955) 267–284. [2] G. Coleman, A. Miller, Electrodialysis of sugar borates, Proc. Iowa Acad. Sci. 49 (1942) 257–261. [3] Y. Hashimoto, M. Itsubiko, K. Mutsuo, Paper electromigration of Xavonoids and sugars using a high constant-voltage current, Nature 170 (1952) 975–976. [4] A.B. Foster, Zone electrophoresis of carbohydrates, Adv. Carbohydr. Chem. Biochem. 12 (1957) 81–115. [5] S. Consden R., W.M. Stanier, Ionophoresis of sugars on paper and some applications to the analysis of protein polysaccharide complexes, Nature 169 (1952) 783–785. [6] J.X. Khym, L.P. Zill, The separation of monosaccharides by ion exchange, J. Am. Chem. Soc. 73 (1951) 2399–2400. [7] J.X. Khym L.P. Zill, The separation of sugars by ion exchange, J. Am. Chem. Soc. 74 (1952) 2090–2094. [8] H.V. Thorne, Electrophoretic studies of components from cells susceptible to poliovirus, Bacteriol. Proc. V6 (1962) 130. [9] H.V. Thorne, Electrophoretic separation of polyoma virus DNA from host cell DNA, Virology 29 (1966) 234–239. [10] E.G. Richards, J.A. Coll, W.B. Gratzer, Disc electrophoresis of ribonucleic acid in polyacrylamide gels, Anal. Biochem. 12 (1965) 452–471. [11] A.C. Peacock, C.W. Dingman, Molecular weight estimation and separation of ribonucleic acid by electrophoresis in agarose–acrylamide composite gels, Biochemistry 7 (1968) 668–674. [12] K. Danna, D. Nathans, SpeciWc cleavage of simian virus 40 DNA by restriction endonuclease of Hemophilus inXuenzae, Proc. Natl. Acad. Sci. USA 68 (1971) 2913–2917. [13] C. Aaij, P. Borst, The gel electrophoresis of DNA, Biochim. Biophys. Acta 269 (1972) 192–200. [14] G.S. Hayward, M.G. Smith, The chromosome of bacteriophage T5.I. Analysis of the single-stranded DNA fragments by agarose gel electrophoresis, J. Mol. Biol. 63 (1972) 383–395. [15] A. Tiselius, ReXections from both sides of the counter, Annu. Rev. Biochem. 37 (1968) 1–24. [16] U.E. Loening, The fractionation of high-molecular-weight ribonucleic acid by polyacrylamide-gel electrophoresis, Biochem. J. 102 (1967) 251–257. [17] P.C. Loewen, H.G. Khorana, Studies on polynucleotides. CXXII. The dodecanucleotide sequence adjoining the C-C-A end of the tyrosine transfer ribonucleic acid gene, J. Biol. Chem. 248 (1973) 3489–3499. [18] T. Maniatis, M. Ptashne, Structure of the lambda operators, Nature 246 (1973) 133–136. [19] W. Gilbert, N. Maizels, A. Maxam, Sequences of controlling regions of the lactose operon, Cold Spring Harbor Symp. Quant. Biol. 38 (1974) 845–855. [20] R. Tsanev, Direct spectrophotometric analysis of ribonucleic acid fractionation by agar-gel electrophoresis, Biochim. Biophys. Acta 103 (1965) 374–382. [21] T. Maniatis, E.F. Fritsch, J. Sambrook, Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, 1982. [22] B. Olivera, P. Baine, N. Davidson, Electrophoresis of nucleic acids, Biopolymers 2 (1964) 245–257. [23] D.H. Flint, R.E. Harrington, Gel electrophoresis of deoxyribonucleic acid, Biochemistry 11 (1972) 4858–4864. [24] F.M. Ausubel (Ed.), Current Protocols in Molecular Biology, Published by Greene Pub. Associates and Wiley-Interscience: J. Wiley, Published by Greene Pub. Associates and Wiley-Interscience: J. Wiley, New York, 1989. [25] J. Sambrook, E.F. Fritsch, T. Maniatis, Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, 1989.

J.R. Brody, S.E. Kern / Analytical Biochemistry 333 (2004) 1–13 [26] B. Shieh, I. Su, W. Chuang, C. Li, Dye-induced denaturation of DNA dissolved in water, Biotechniques 24 (1998) 60–62. [27] H. Jordan, Gel loading buVer aVects electrophoresis results, Focus 19 (1997) 60–61. [28] D. Pisa-Williamson, C. Fuller, Glycerol-tolerant DNA sequencing gels, US Biochem. Comments 19 (1992) 29–36. [29] W.Y. Ha, P.C. Shaw, J. Wang, Improved electrophoretic resolution of DNA fragments in samples containing high concentrations of salts, Biotechniques 26 (1999) 425–426. [30] H. ErXe, R. Ventzki, H. Voss, S. Rechmann, V. Benes, J. Stegemann, W. Ansorge, Simultaneous loading of 200 sample lanes for DNA sequencing on vertical and horizontal, standard and ultrathin gels, Nucleic Acids Res. 25 (1997) 2229–2230. [31] S. Hjerten, Dedication to professor arne tiselius, Ann. NY Acad. Sci. 209 (1973) 5–7. [32] L.E. Kay, Laboratory technology and biological knowledge: the Tiselius electrophoresis apparatus, 1930–1945, Hist. Philos. Life Sci. 10 (1988) 51–72. [33] R.C. Allen, B. Budowle, Gel Electrophoresis of Proteins and Nucleic Acids: Selected Techniques, de Gruyter, Berlin, New York, 1994. [34] D.M. Hawcroft, Electrophoresis, IRL Press at Oxford University Press, Oxford, New York, 1997. [35] J.R. Brody, S.E. Kern, Sodium boric acid: a Tris-free, cooler conductive medium for DNA electrophoresis, Biotechniques 36 (2004) 214–216. [36] J. Sambrook, D.W. Russell, Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 2001. [37] P. McPhie, J. Hounsell, W.B. Gratzer, The speciWc cleavage of yeast ribosomal ribonucleic acid with nucleases, Biochemistry 5 (1966) 988–993. [38] N.C. Stellwagen, A. Bossi, C. GelW, P.G. Righetti, DNA and buVers: are there any noninteracting, neutral pH buVers?, Anal. Biochem. 287 (2000) 167–175. [39] Q. Liu, X. Li, S.S. Sommer, pK-matched running buVers for gel electrophoresis, Anal. Biochem. 270 (1999) 112–122. [40] U.E. Loening, Fractionation of low-molecular-weight ribonucleic acid, Biochem. J. 114 (1969) 25P–26P. [41] N.C. Stellwagen, C. GelW, P.G. Righetti, DNA–histidine complex formation in isoelectric histidine buVers, J. Chromatogr. A 838 (1999) 179–189. [42] N.C. Stellwagen, Agarose gel pore radii are not dependent on the casting buVer, Electrophoresis 13 (1992) 601–603. [43] Y. Kukita, T. Tahira, S.S. Sommer, K. Hayashi, SSCP analysis of long DNA fragments in low pH gel, Hum. Mutat. 10 (1997) 400– 407. [44] D.G. McLaren, O. Boulat, D.D. Chen, Univalent salts as modiWers in micellar capillary electrophoresis, Electrophoresis 23 (2002) 1912–1920. [45] E. Stellwagen, N.C. Stellwagen, The free solution mobility of DNA in Tris–acetate–EDTA buVers of diVerent concentrations, with and without added NaCl, Electrophoresis 23 (2002) 1935– 1941.

13

[46] A. Gordon, B. Keil, K. Sehesta, F. Knessel, An electrophoretic investigation of the mixture of oligonucleotides formed by the enzymatic degradation of deoxyribonucleic acid by deoxyribonuclease, J. Biochem. (Tokyo) 48 (1951) 569–575. [47] M. Spencer, Anomalous conductivity zones in electrophoresis I. Basic theory for two-ion systems, Electrophoresis 4 (1983) 36–41. [48] O. Bilenko, D. Gavrilov, B. Gorbovitski, V. GorWnkel, M. Gouzman, G. Gudkov, V. Khozikov, O. Khozikov, O. Kosobokova, N. Lifshitz, S. Luryi, A. Stepoukhovitch, M. Tcherevishinick, G. Tyshko, Formation of a resistive region at the anode end in DNA capillary electrophoresis, Electrophoresis 24 (2003) 1176–1183. [49] D. Figeys, H. Ahmadzedeh, E. Arriaga, N.J. Dovichi, Pseudo-coulometric loading in capillary electrophoresis DNA sequencing, J. Chromatogr. A 744 (1996) 325–331. [50] P. Levene, L. Bass, H. Simms, The ionization of pyrimidines in relation to the structure of pyrimidine nucleosides, J. Biol. Chem. 70 (1926) 229–241. [51] P. Levene, H. Simms, The dissociation constants of plant nucleotides and nucleosides and their relation to nucleic acid structure, J. Biol. Chem. 65 (1925) 519–534. [52] P. Levene, H. Simms, L. Bass, The eVect of ionization upon optical rotation of nucleic acid derivatives, J. Biol. Chem. 70 (1926) 243– 251. [53] P.H. Barry, J.W. Lynch, Liquid junction potentials and small cell eVects in patch–clamp analysis, J. Membr. Biol. 121 (1991) 101– 117. [54] B. Ng, P.H. Barry, The measurement of ionic conductivities and mobilities of certain less common organic ions needed for junction potential corrections in electrophysiology, J. Neurosci. Methods 56 (1995) 37–41. [55] R. Weinberger, Practical Capillary Electrophoresis, Academic Press, San Diego, CA, 2000. [56] J. Böeseken, The use of boric acid for the determination of the conWguration of carbohydrates, Adv. Carbohydr. Chem. Biochem. 4 (1949) 189–210. [57] N. Stellwagen, C. GelW, P.G. Righetti, The use of gel and capillary electrophoresis to investigate some of the fundamental physical properties of DNA, Electrophoresis 23 (2002) 167–175. [58] B. Vogelstein, D. Gillespie, Preparative and analytical puriWcation of DNA from agarose, Proc. Natl. Acad. Sci. USA 76 (1979) 615– 619. [59] B. Kozulic, Looking at bands from another side, Anal. Biochem. 216 (1994) 253–261. [60] A. Olsson, T. Moks, M. Uhlen, A. Gaal, Uniformly spaced banding pattern in DNA sequencing gels by use of Weld-strength gradient, J. Biochem. Biophys. Methods 10 (1984) 83–90. [61] U.E. Loening, J. Ingle, Diversity of RNA components in green plant tissues, Nature 215 (1967) 363–367. [62] T. Ray, A. Mills, P. Dyson, Tris-dependent oxidative DNA strand scission during electrophoresis, Electrophoresis 16 (1995) 888–894. [63] R. West, The electrophoretic mobility of DNA in agarose gel as a function of temperature, Biopolymers 26 (1987) 607–608. [64] R.S. Tang, D.E. Draper, Bulge loops used to measure the helical twist of RNA in solution, Biochemistry 29 (1990) 5232–5237.